Reference line approach for vector data compression

A lossy compression algorithm for vector data based on vector quantization with preliminary polygonal approximation is considered. The main idea of the proposed approach is the use of reference lines to reduce redundancy of input vector data. The references lines are constructed as coarse polygonal approximation of input curves and compressed by a lossless algorithm. The residual vectors are then encoded by vector quantization. The proposed algorithm achieves a better rate-distortion performance than the previous cluster-based algorithm. The achieved results are close to the optimized single-level DPCM modeling scheme.